Alex Fink wrote:
> On Sat, 27 Aug 2016 18:48:48 -0400, Herman Miller
> <[log in to unmask]> wrote:
>> Instead, my current idea is that there were at least 5 vowels /a e
>> i o u/ and possibly a central vowel /ə/ or /ɨ/. Most of the /e/ and
>> /o/ merged with /i/ and /u/, but they remained as /e/ and /o/ in
>> certain environments. Diphthongs /ei/ and /ou/ changed to long
>> vowels /iː/ and /uː/. This accounts for the overabundance of /i/
>> and /u/ in comparison with /e/ and /o/, and the rarity of /e/ and
>> /o/ at the end of words.
> I guess it makes sense that the central vowels fail to participate in
> these changes, as it is peripheral vowels that have the tendency to
> rise.  But where do modern Tirëlat /@/ and /1/ come from, then?  It
> sounds like you want at most one of those to be a survival.

The history of the central vowels is one of those things I haven't 
worked out. At first I was considering that unstressed /ə/ was just 
lost, and some /ə/ may have merged with /a/. The current idea I'm trying 
out is that a single central vowel may have split into modern /ɜ/ and 
/ɨ/, the same kind of split that happened with /e/ and /o/ but without 
the merger.

>> My current working assumption is that /e/ and /o/ remained /e/ and
>> /o/ in closed syllables, and merged with /i/ and /u/ elsewhere.
>> Word-final /ə/ was lost.
> In which order?

The merger happened before the loss of word-final /ə/. So a word like 
"mek" was originally *mek, but "mik" could be from *mik, *mekə, or *mikə.

>> Where modern Tirëlat has /e/ and /o/ in open syllables, there may
>> have been sound changes that eliminated one of the consonants, such
>> as assimilation or degemination. Examples:
>> *veddi > vedi "narrow" *glemma > glema "blurry" *merlo > melu "sea
>> serpent"
>> One possibility is that the cluster /ts/ was distinct from the
>> affricate that became /ts/ in modern Tirëlat (I'll write that as
>> /c/).
>> *keci > kitsi "claw" *ketsi > ketsi "binary"
> Reasonable, but the examples are on the conservative side.  I forget
> whether I've already asked this, but what sort of sources, if any,
> might Tirëlat have adopted loans from?  That's an easy way to re-fill
> phonotactic gaps created by sound change, which you might otherwise
> be hesitant to posit because modern Tirëlat clusters are too diverse
> and you can't lose any.  Or could there be something going on like
> internal syncope, or re-generalisation of the most common allomorph
> of a derivational suffix?

Modern Tirëlat could adopt words from pretty much any other language 
spoken in the same world (including Jarda and Lindiga); these being 
relatively recent sound changes. Lindiga does have central vowels.

Syncope may be involved in the word "mërimri" at least; it was probably 
"mërimëri" at some point in the past. And certainly words like "tkwyr" 
are likely to have lost vowels at some point.

> Did pre-Tirëlat have vowel-initial suffixes to consonant-final bases?
> That would create alternations in syllable openness which would then
> manifest as height alternations.  What happens to these?  Levelling?
> Any doublets?
> Alex

That's a good point; the diminutive of "žeğ" ought to be "žiğits" if it 
follows the pattern, and the verbal noun of "tos" should be "tusat". But 
I've already used "žeğits".